Weird repeating noise pattern in DSLR stacked image

[rnobleeddy]

Had a clear spell last night so gave an old Russian 300mm lens a go with my DSLR. Overall, can't complain - it's got better stars in the corners than my £400 60mm APO refractor!

However, when I stretched the image and started processing, I spotted a weird repeating pattern of noise. It's all over the image - I highlighted a few below. 

Does anyone know what this is? 

It's 25x300s subs. Dither every sub. Stacked in DSS with (as far as I checked) the settings recommended by startools.

As best I can tell these are exact repeating patterns - so either camera or calibration or stacking?

Stacked, stretched image

 

Single sub, stretched

 

 

 

 

 

[bobro]

Could be hot pixels in different places due to dithering, though I can't see any in the single sub. What stacking method was used? (Kappa-sigma works well in eliminating hot/cold pixels when dithering is used.) 

[rnobleeddy]

Just now, bobro said:

Could be hot pixels in different places due to dithering, though I can't see any in the single sub. What stacking method was used? (Kappa-sigma works well in eliminating hot/cold pixels when dithering is used.) 

Thanks. I was using kappa-sigma. Dithers weren't so large that he same bit of the sensor could explain all of these. 

I had messed up the flats (overexposed) but whilst fixing those improved the image, it didn't fix these.

What did was turning on hot/cold pixel detection in DSS. I now have

 

 

So still a little unsure what caused this. My dithers are probably less aggressive than recommended but DSS is reporting offsets of >100 pixels on some subs. So I'm still leaning towards some weird stacking artifact or some other calibration issue, that just happens to be taken care of by the hot pixel removal. 

[alacant]

Hi

That's what i'd expect when not stacked with a clipping algorithm. A rough hot pixel count shows ~25 dots, corresponding to the number of frames. 

I'd increase the dither to around 15 pixels and check you really are using a clipping algorithm when stacking, e.g. Linear clip 5-4 works well for our dslrs even with the most stubborn satellite trails. Not sure what app you're using for capture but remember that although the dither is normally set for the guide camera, it must be converted to give the required movement on the imaging camera.

Cheers and HTH

[rnobleeddy]

6 hours ago, alacant said:

Hi

That's what i'd expect when not stacked with a clipping algorithm. A rough hot pixel count shows ~25 dots, corresponding to the number of frames. 

I'd increase the dither to around 15 pixels and check you really are using a clipping algorithm when stacking, e.g. Linear clip 5-4 works well for our dslrs even with the most stubborn satellite trails. Not sure what app you're using for capture but remember that although the dither is normally set for the guide camera, it must be converted to give the required movement on the imaging camera.

Cheers and HTH

Aha - that starts to make sense. My brain hadn't really processed it properly. So it's a series of randomly distributed hot pixels and the repeating pattern is the dither pattern. 

I can't say for sure, but when I checked back to DSS just now, it was using average rather than the cliping algorithm. So let's assume that was the issue! I was perhaps too tired to do anything useful last night.

Next up to fix is what appears to be horrendous tilt! The centre of the frame was in focus, but either corner is out of focus, one corner with a red halo, one corner with blue

 

Edited October 1, 2021 by rnobleeddy

[alacant]

16 minutes ago, rnobleeddy said:

series of randomly distributed hot pixels

The rogue pixels are found at fixed positions on the camera sensor. 

Each group of 25 points is produced by the same hot pixel. Due to the dither, aligning the stars in each frame prior to stacking causes the pixel to appear to move. Of course, it remains at exactly the same coordinates on the camera's sensor.

 

 

Edited October 1, 2021 by alacant

[rnobleeddy]

5 minutes ago, alacant said:

The rogue pixels are found at fixed positions on the camera sensor. 

Each group of 25 points is produced by the same hot pixel. Due to the dither, aligning the stars in each frame prior to stacking causes the pixel to appear to move. Of course, it remains at exactly the same coordinates on the camera's sensor.

 

 

Sorry - that's what I meant - I couldn't understand why the 25 pixels appeared to repeat at random intervals - so some groups were close together or overlapping, and then some of the frame looked ok.

I guess I don't really mean the hot pixels are randomly located - the pattern is random, but once we know it, it's the same on each sub!